Date: Tue, 7 Mar 95 00:34 WET From: pete@pelican.com (Pete Carah) To: current@FreeBSD.org Subject: Re: more ETXTBSY bugs Message-ID: <m0rlujL-000K2tC@pelican.com> In-Reply-To: <9503062157.AA19959@cs.weber.edu>
next in thread | previous in thread | raw e-mail | index | archive | help
In article <9503062157.AA19959@cs.weber.edu> Terry writes:
.....
>I'd like to see the people seeing the thrashing problems add about 16M
>of swap without doing any extra apllications and see what happens. If
>the vnode pager is happy, this should be doable with a swap on file.
>My immediate suspicion is that the problems would just go away until
>the cache was filled again (this time consuming all of swap).
I see NO swap usage to speak of; this problem doesn't appear to have
much if anything to do with swap. System has 32mb ram and a 486DX2/66.
disks are 2x345ide and 1x213scsi (all maxtor if it matters; scsi
is a 1742 with some letter). 32mb ram is 8192 pages less the 384k hole;
that should be 8096 pages if I figure right.
I can run netscape during this thrashing with *no* apparent effect on
either (though netscape's performance via a proxy connected at 14.4k
is slow at best...) It was not running for the following; only the
make world (which is about at make all in usr.sbin/pkg_install/lib)
and one csh to look with.
Following taken _during_ (two different) seek-chatter during a make world:
------------------------------------------------------------------
puffin% swapinfo
Device 1K-blocks Used Available Capacity
/dev/wd0b 36864 3800 33064 10%
/dev/wd1b 0 *** not available for swapping ***
/dev/sd0b 0 *** not available for swapping ***
/dev/sd1b 0 *** not available for swapping ***
puffin% vmstat -s
4297752 cpu context switches
40451654 device interrupts
1948048 software interrupts
13828356 traps
13166711 system calls
72 swap pager pageins
157 swap pager pages paged in
437 swap pager pageouts
785 swap pager pages paged out
2621 vnode pager pageins
7694 vnode pager pages paged in
4 vnode pager pageouts
4 vnode pager pages paged out
2989035 VM object cache lookups
2662043 VM object hits
89 page daemon wakeups
237102 pages examined by the page daemon
0 pages reactivated
467 intransit blocking page faults
153713529 zero fill pages allocated
5667612 zero fill pages zeroed
1937306 copy-on-write faults
14124043 total VM faults taken
14163547 pages freed
0 pages freed by daemon
6746415 pages freed by exiting processes
3650 pages active
856 pages inactive
1603 pages in VM cache
992 pages wired down
505 pages free
4096 bytes per page
13160106 total name lookups
cache hits (56% pos + 3% neg) system 0% per-process
deletions 0%, falsehits 10%, toolong 0%
-----------------------------------------------------------------------
The source tree is on scsi and everything else (/usr including obj) is
on wd0. At least half of the disk activity is on the scsi, so couldn't
be swap.
-- Pete
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?m0rlujL-000K2tC>